Methods of single cell genotyping and uses thereof

Abstract

The present disclosure relates to methods for generating single cell transcriptomic and genotyping models and data structures thereof, including uses thereof with machine-learning algorithms, e.g., for monitoring disease progression, diagnosis, and treatment selection.

Description

[0001] METHODS OF SINGLE CELL GENOTYPING AND USES THEREOF

[0002] FIELD

[0003] The present disclosure relates to methods of genotyping single cells and uses thereof, e.g., for monitoring disease progression and treatment selection in subjects.

[0004] CROSS-REFERENCE TO RELATED APPLICATIONS

[0005] The present application claims priority to and benefit from U.S. Provisional Patent Application No. 63 / 585,793, filed September 27, 2023, and U.S. Provisional Patent Application No. 63 / 627,428, filed January 31, 2024, the entirety of each of which is incorporated by reference herein.

[0006] BACKGROUND

[0007] In many diseases, such as hematopoietic cancers and neoplasms, disease progression can be driven by a small number of somatic mutations in protein-coding and non-protein coding genes. Transcriptional profiling of mutation-bearing clonal subpopulations can provide significant insights into these malignancies. However, successful transcriptional profiling requires simultaneous profiling of gene expression and determining the cellular genotype at specific loci in individual cells within a heterogenous population.

[0008] Current single-cell genotyping approaches are limited to a single locus and require development of a complex series of amplification primers for each target. There remains a need to improve single cell resolution of transcriptome analysis and leverage the data into the clinic.

[0009] SUMMARY

[0010] Accordingly, in one aspect, the disclosure provides methods that address and overcome the limitations of single cell genotyping and its utility, e.g., in identifying disease-causing mutations, measuring progression of disease, and / or treatment.

[0011] In aspects, the disclosure provides a method for single cell genotyping. In embodiments, the methods comprise generating a single cell gene expression profile from a single cell (e.g., originating from a sample of cells and / or a clonal cell population), where the single cell gene expression profile comprises sequencing data from a cDNA library and / or gene capture. In embodiments, the methods comprise performing a machine-learning process using the sequencing data from the cDNA library and / or the gene capture to generate a single cell gene expression model. In embodiments, the single cell gene expression model is used to provide information for annotating a single genotype. In embodiments, the method comprises refining the single cell gene expression model using the sequencing data from the cDNA library or the gene capture. In embodiments, the method comprises annotating one or more single cell genotypes using the single cell gene expression model.

[0012] In embodiments, the method comprises obtaining a sample of cells, for example from a blood draw or biopsy. In embodiments, methods herein comprise expanding one or more clonal cell populations from one or more cells originating from the sample.

[0013] In embodiments, the methods provided herein comprise generating single cell gene expression profile. In embodiments, the methods provided herein comprise generating cDNA library from a single cell (e.g., comprising performing reverse transcriptase PCR (RT-PCR) from mRNA).

[0014] In embodiments, generating the cDNA library comprises barcoding the cDNA, e.g., where barcoding further comprises performing one or more of fragmentation, end repairing, A-tailing, and adaptor ligation, optionally performing each of fragmentation, end repairing, A-tailing, and adaptor ligation. In embodiments, methods comprise amplifying the cDNA prior to and / or during barcoding, e.g., using PCR amplification techniques.

[0015] In embodiments, generating the single cell gene expression profile comprises sequencing the cDNA library, optionally the barcoded cDNA library (e.g., using any method described herein, such as performing one or more of nested PCR; short read sequencing, long read sequencing, or both; amplicon sequencing; single-cell RNA sequencing (scRNA-seq), scTag-seq, single-cell transposase-accessible chromatin sequencing (scATAC-seq), CyTOF / SCoP, E-MS / Abseq, miRNA-seq, CITE-seq, and combinations thereof). In embodiments, sequencing the barcoded cDNA library comprises indexing the barcoded cDNA library using sample index PCR, wherein each output provides a barcoded DNA sequence.

[0016] In embodiments, the methods herein comprise performing whole genome sequencing (WGS) to generate genomic sequencing data, e.g., from a single cell from a clonal cell population. WGS data, in embodiments, is used to generate one or more lineage trees of one or more genes of interest. In embodiments, methods comprise performing a machine-learning process using the WGS data, to generate a single cell gene expression model and / or to refine the single cell gene expression model.

[0017] In embodiments, the methods described herein comprise performing single cell gene capture (e.g., “hybrid capture”) to generate gene capture sequence data of one or more genes of interest. In embodiments, gene capture is performed with the cDNA, and comprises hybridizing the cDNA with one or more hybridization probes (e.g., TWIST hybridization probes) that target one or more of the genes of interest. In embodiments, the one or more hybridization probes and / or one or more adaptors are chemically-modified, e.g., biotinylated, streptavidin-tagged, etc., to enable enrichment of one or more of the genes of interest using beads, plates, well, etc. In embodiments, the enrichment includes using magnetic beads to bind the one or more hybridizing probes and / or adaptors via a biotin-streptavidin interaction, where the magnetic beads are pulled down by applying a magnetic field.

[0018] In embodiments, generating the gene capture sequence data comprises performing one or more of sample index PCR and / or nested PCR (e.g., for amplification), and / or sequencing one or more of the genes of interest using short read sequencing, long read sequencing, or both. In embodiments, generating the gene capture sequence data comprises amplicon sequencing. In embodiments, generating the gene capture sequence data comprises performing one or more of single-cell RNA sequencing (scRNA-seq), scTag-seq, single-cell transposase-accessible chromatin sequencing (scATAC-seq), CyTOF / SCoP, E-MS / Abseq, miRNA-seq, CITE-seq, and combinations thereof.

[0019] In embodiments, the methods herein comprise performing one or more machine-learning processes, e.g., including manifold learning and / or dimensionality reduction. In embodiments, manifold learning comprises performing one or more of uniform manifold approximation and projection (UMAP), locally linear embedding, local linear isometric mapping (ISOMAP), t- distributed stochastic neighbor embedding (t-SNE), or Potential of Heat-Diffusion for Affinity Based Trajectory Embedding (PHATE). In embodiments, dimensionality reduction comprises performing one or more of uniform manifold approximation and projection (UMAP) algorithm, principal component analysis (PCA), random projection, independent component analysis, feature selection method, factor analysis, Sammon mapping, curvilinear components analysis, stochastic neighbor embedding (SNE), local linear isometric mapping (ISOMAP), maximum variance unfolding, locally linear embedding, t-distributed stochastic neighbor embedding (t-SNE), nonnegative matrix factorization, kernel principal component analysis, graph-based kernel principal component analysis, linear discriminant analysis, a generalized discriminant analysis, LargeVis, Laplacian Eigenmap, and Fisher's linear discriminant analysis. In embodiments, methods herein comprise performing one or more clustering machine-learning processes, e.g., using HDBSCAN and / or Louvain community detection.

[0020] In embodiments, methods herein comprise annotating one or more single cell genotypes using the single cell gene expression model. In embodiments, annotating includes identifying one or more mutations and / or single nucleotide polymorphisms, e.g., comprising one or more substitutions, variants, deletions, duplications, insertions, translocations, or indels. Methods herein, in embodiments, detect somatic mutations and / or germ line mutations. In embodiments, the one or more mutations is located in one or more of non-protein coding RNAs, intronic sequences, promoters, enhancers, silencers, insulators, structural elements of chromosomes, transcription-factor binding elements, histone-binding regions, telomeres, and satellite DNA. In embodiments, the one or more mutations cause or are correlated with one or more changes in protein function (e.g., such as the JAK / STAT signaling pathway).

[0021] In embodiments, annotating the single cell genotype comprises classifying the single cell into one or more categories (e.g., a healthy (wild-type) category and a diseased (abnormal) category) based on the identification of one or more mutations.

[0022] In embodiments, methods herein the improve the detection of a gene signature compared to detection of a gene signature using single cell genotyping alone or compared to using single cell gene expression alone.

[0023] In aspects, the present disclosure is directed to methods of single cell genotyping comprising obtaining a sample of cells and expanding one or more clonal cell populations from the sample of cells, generating a single cell gene expression profde from a single cell in the sample of cell and / or from the one or more clonal cell populations, wherein the single cell gene expression profile comprises sequencing data from a cDNA library and gene capture, generating genomic sequencing data by performing whole genome sequencing (WGS) using a single cell from the one or more clonal cell populations and generating one or more lineage trees using the genomic sequencing data, performing a machine-learning process using one or more of the sequencing data from the cDNA library, gene capture, and the genomic sequencing data to generate a single cell gene expression model, refining the single cell gene expression model using one or more of the sequencing data from the cDNA library, gene capture, and the genomic sequencing data, and annotating one or more single cell genotypes using the refined single cell gene expression model, wherein annotating comprises identifying one or more mutations in the single cell, and classifying the single cell into one or more categories as a function of the identifying.

[0024] In aspects, the present disclosure is directed to methods for measuring disease progression using any of the methods of genotyping described herein, where at least a first single cell genotype is annotated at a first timepoint and at least a second single cell genotype is annotated at a second timepoint, wherein a difference in one or more mutations between the first timepoint and second timepoint indicates a progression in disease state.

[0025] In embodiments, the single cell originates from a blood draw, biopsy, or a clonal cell population derived therefrom, optionally wherein the blood draw comprises circulating tumor cells, or the biopsy is a tumor biopsy.

[0026] In embodiments, a time between the first timepoint and the second timepoint is about or at least about 1 day, about or at least about 2 days, about or at least about 3 days, about or at least about 1 week, about or at least about 2 weeks, about or at least about 3 weeks, about or at least about 1 month, about or at least about 2 months, about or at least about 6 months, about or at least about 1 year, about or at least about 2 years, or about or at least about 5 years.

[0027] In embodiments, the methods described herein further comprise stratifying a subject as a function of the progression in disease state, optionally wherein stratifying comprising classifying the subject into one or more categories based on a change in the disease state. In embodiments, the one or more categories relate to one or more diseases, optionally comprising a solid tumor, hematological cancer, acquired disease, congenital disease, hematopoietic disease, non- hematopoietic disease, disease affecting bone marrow, neurons, astrocytes, oligodendrocytes, cardiomyocytes, skeletal muscle cells, hepatocytes, pancreatic 0 cells, and lung epithelial cells.

[0028] In aspects, the present disclosure is directed to methods for treating a disease in a subject in need thereof using any of the methods of genotyping described herein, where the method informs the selection of a treatment, adjustment of one or more doses of a treatment, and / or the termination of a treatment. In embodiments, the methods described herein comprise performing a correlation activity analysis to identify a compound that induces a transcriptional signature. In embodiments, the methods described herein comprise performing a leave-one-out classifier analysis to identify a compound-specific differential transcriptional signature. In embodiments, the methods described herein comprise performing a nearest-neighbor analysis to assess the consistency and / or reproducibility of compound-specific differential transcriptional signatures. In embodiments, the methods described herein comprise performing a differential expression analysis to determine which genes are differentially regulated by one or more compounds.

[0029] In embodiments, the disease is or relates to one or more of a solid tumor, hematological cancer, acquired disease, congenital disease, hematopoietic disease, non-hematopoietic disease, disease affecting bone marrow, neurons, astrocytes, oligodendrocytes, cardiomyocytes, skeletal muscle cells, hepatocytes, pancreatic P cells, and lung epithelial cells.

[0030] In embodiments, the disease is a solid tumor or hematological cancer, e.g., where the single cell genotype originates from a blood draw, biopsy, or clonal cell population obtained from the subject, optionally wherein the blood draw comprises circulating tumor cells, or the biopsy is a tumor biopsy. In embodiments, the subject is undergoing treatment for cancer. In embodiments, the subject is treatment refractory or has a disease that has developed drug resistance, and / or wherein the method identifies that the subject is treatment refractory or has a disease that has developed drug resistance. In embodiments, the treatment comprises one or more of an immunotherapy, chemotherapy agent, radiation therapy, and surgery.

[0031] In aspects, described herein are libraries of cellular data, e.g., generated from the method described herein, wherein the library of cellular data comprises one or more single cell genotypes matched to one or more disease states, and / or data relating to the identities of one or more mutations and / or single nucleotide polymorphisms in one or more single cell genotypes, and / or data relating the one or more mutations to one or more disease states, one or more mutations and / or single nucleotide polymorphisms that change as a function of time and / or kinetics of mutation change, one or more whole genome sequences, and one or more machine-learning models and / or outputs for annotating single cell genotypes.

[0032] In one aspect, the disclosure provides a method for measuring disease progression by analyzing single cell genotype information generated by the methods described herein. BRIEF DESCRIPTION OF FIGURES

[0033] FIG. 1 depicts a diagrammatic representation of the results of a hybrid capture panel design in comparison between a 17-gene panel and a 2-gene panel, which shows fractional percent of cells in a sample and fold-expression changes relative to the group.

[0034] FIG. 2A - 2C depicts a diagrammatic representation of the results of clustering single cell genotypes from cell lines with known mutation status. FIG. 2A shows a uniform manifold approximation and projection (UMAP) visualization of UKE-1 and HL60 clustering, with identified mutations plotted. FIG. 2B shows individually-identified sequences from nested PCR Seq of each cell line, delineated as wild-type, mutant, or ambiguous. FIG. 2C shows the refined uniform manifold approximation and projection (UMAP) visualization based on long-read sequencing, which resolves several ambiguous data points.

[0035] FIG. 3 depicts a diagrammatic representation of the results of single cell genotyping compared across nested PCR and long-read sequencing with homogenous and mixed cell samples of UKE-1 and HL-60 cell lines.

[0036] FIG. 4 depicts a diagrammatic representation of the results of single cell genotyping to identify healthy and disease subpopulations based on identifiable mutant genotypes.

[0037] DETAILED DESCRIPTION

[0038] In many diseases, such as, by way of non-limitation, hematopoietic cancers and neoplasms, disease progression is driven by a small number of somatic mutations in both non-protein-coding and protein-coding genes. Transcriptional profiling of mutation-bearing clonal subpopulations provides significant insights into these diseases. In embodiments, this requires simultaneous profiling of gene expression and determination of the genotype of specific loci in individual cells within a heterogenous population. Current single-cell genotyping approaches are limited to a single locus and require development of a complex series of amplification primers for each target.

[0039] One major disadvantage of current single cell RNA sequencing is the loss of histological information as both single cell and single nuclei suspensions are generally prepared from tissues. Although trajectory analysis (e.g., using UMAP) can help with projecting the association and transition between different cell types, other confounding factors associated with tissue digestion, cell isolation, and preservation could alter gene expression and cell representation. Additionally, dimensionality reduction techniques for visualizing single cell expression analyses are only 2- dimensional.

[0040] In aspects, the disclosure provides methods for single cell genotyping which improve upon standard methods. In embodiments, the methods comprise using a single cell from a sample, or a single cell from a clonal cell population (e.g., enriched from the sample), and generating both a transcriptome profile and a genotype sequence from a barcoded and pooled cDNA library. In embodiments, the barcoded and pooled cDNA library is apportioned and prepared by the standard 10X Chromium method (10X GENOMICS, microfluidic chip-based single cell droplet cDNA barcoding, amplification, sequencing, and transcriptional profiling). In embodiments, these sequencing results are used as inputs / training data with a machine-learning process to train, output, or refine a model (e.g. , UMAP, t-SNE, or PC A), which provides a 2-dimensional or 3-dimensional map / plot that clusters genetic / expression characteristics, where the distance between clusters relates to one or more features, e.g., similarity or dissimilarity. In embodiments, the cDNA library is subjected to gene capture using Twist hybridization probes coupled with magnetic bead capture (TWIST BIOSCIENCES, next-generation sequencing (NGS) target enrichment). In embodiments, the enriched sequencing data (e.g., long-read sequencing data) is used as an input to train, output, or refine a model. In embodiments, machine-learning models are used to annotate single cell genotypes, identifying a variety of mutations (e.g., substitutions, variants, deletions, duplications, insertions, translocations, and / or indels) and / or single nucleotide polymorphisms.

[0041] The model refinement, in embodiments, results in a visualization which has increased dimensionality (e.g., 3-dimension visualization of data), where the 2-D visualization (x-, y- axes) shows categorical clustering (e.g., disease vs healthy / wild-type; before vs. after treatment, etc.), where each individual point in the clustering represents a wild-type / mutant sequence (e.g., gene sequence, transcript, and / or expressed protein), coupled with a third dimension (z-axis) which communicates amplitude / magnitude information (e.g., gene expression level, relative abundance, etc.).

[0042] In embodiments, the methods described herein for single cell genotyping combine single cell transcriptome analysis with dimensionality reduction techniques to visualize whole cell molecular dynamics with higher resolution than standard methods. Cells

[0043] In aspects, the methods described herein include in vitro or ex vivo analyses of single cells obtained from heterogeneous samples (e.g., blood, biopsy, tissue resection, etc.), or obtained from homogenous samples (e.g., a clonal cell population originating from a single cell isolated from a subject, sample, a cell line, etc.).

[0044] In embodiments, the cell is a circulating tumor cell, for example, obtained from a sample originating from a subject (e.g., blood draw, bone marrow biopsy, lymph, tumor biopsy, etc.). In embodiments, the cell is obtained from a biopsy, including a tumor biopsy (e.g., resurrected from a primary tumor, secondary tumor, etc.), or a tissue biopsy (e.g., a lymph node, diseased organ, tissue suspected to be diseased, etc.). In embodiments, the subject has metastases, or is believed to have a metastatic cancer. In embodiments, the subject is healthy, or otherwise not expected to have a disease at the time a sample is obtained.

[0045] In embodiments, the cell belongs to a cell line, including commercially-available cell lines and cell stocks. The cell line, in embodiments, is a human cell line, including for example, HL60, UKE-1, HEK (e.g., HEK293, 293T, and derivatives), HeLa, MCF-7, HT-1080, PER.C6, MSC-1, HEPG2, TUP-1, A549, XCL-1, KG-1, MEC, and LS123 cell lines. In embodiments, the methods described herein are suitable with any cell type or cell organization, including in non-limiting examples, virally-infected cells, non-human cells, adherent and non-adherent cells, cells configured for 3D cell culture, and organoids.

[0046] In embodiments, the present methods of single cell genotyping find use in validating and / or annotating stem cell lines (e.g., in creating cell banks and / or libraries thereof), and / or for selecting clonal cell populations for manufacturing cell therapies and / or for differentiation into cell banks.

[0047] In such embodiments, the cell comprises a stem cell, progenitor cell, or pluripotent stem cell, which includes an induced pluripotent stem cell (iPSC). In embodiments, these cells are derived from peripheral blood, cord blood, bone marrow, tissue, or organs. In embodiments, the cell population comprises cells that are isolated from blood, including without limitation, immune cells, red blood cells, platelets, and hematopoietic cell lineages or precursors thereof, such as common lymphoid precursor (CLP) cells, granulocyte-monocyte progenitor (GMP) cells, progenitor-T cells, T lymphocytes, B lymphocytes, Natural Killer cells, neutrophils, dendritic cells, monocyte, macrophages, red cells, megakaryocytes, and platelets. In embodiments, the cell comprises a non-hematopoietic cell population, including, but not limited to, mesenchymal stem cells, epithelial stem cells, or neural stem cells. In embodiments, the non-hematopoietic cell is selected from neurons, astrocytes, oligodendrocytes, cardiomyocytes, skeletal muscle cells, hepatocytes, pancreatic P cells, lung epithelial cells, and progenitors thereof.

[0048] In embodiments, the cell is obtained from a clonal expansion of one or more cells obtained from a larger pool of cells. In embodiments, the cell is obtained from cell culture or fresh frozen tissue, such as formalin-fixed paraffin-embedded (FFPE) tissue. Persons skilled in the art, with the benefit of this disclosure in its entirety, will be aware of the various methods for clonal cell expansion and isolation, as well as tissue handling methods, that are compatible with the methods described herein.

[0049] Methods of Single Cell Transcriptomics and Genotyping

[0050] In aspects, described herein are methods of generating a single cell gene expression profile and genotyping a single cell. This includes, in embodiments, generating a cDNA library from a single cell. In embodiments, methods of generating a cDNA library include one or more steps of barcoding, fragmentation, end repair, A-tailing, adaptor ligation, sample index PCR, and sequencing, e.g., incorporating one or more commercially-available techniques or protocols, for example, from 10XGENOMICS, ILLUMINA, TWIST BIOSCIENCES, etc. In embodiments, methods of generating a single cell gene expression profile and genotyping a single cell comprise PCR amplification methods, e.g., using AMPLISEQ (ILLUMINA), which generates DNA and / or cDNA amplicons having primer sequences at the 5’ and / or 3’ ends, removing the primers (e.g, using fragmentation) from the amplicon ends, ligating adaptors to the 5’ and / or 3’ ends of the amplicons, PCT amplifying the amplicons with the adapters, and sequencing. In embodiments, sequencing data (e.g., originating from cellular DNA and / or mRNA) is then transformed through the use of machine learning ( .g., UMAP projection) into a form that is useful for extracting mutational and genotyping information with single cell resolution.

[0051] In embodiments, the methods described herein include deriving single cell transcriptomic and whole-genome sequencing to identify mutations, even before the manifestation of a disease phenotype. In embodiments, a sample is obtained form a subject, where cells are sorted / isolated from the sample based on one or more characteristics. In embodiments, one or more cells are expanded (e.g., clonal cell expansion), which are used for whole genome sequencing (WGS) to generate genomic sequencing data. In embodiments, genomic sequencing data is used as input data to generate one or more lineage trees (e.g., for each gene). Illustrative phylogenetic dynamics inference algorithm for generating lineage trees are known in the art, for example, as described in Van Egeren et al., “Reconstructing the Lineage Histories and Differentiation Trajectories of Individual Cancer Cells in Myeloproliferative Neoplasms,” Cell Stem Cell. 2021 Vol. 28, No. 3, 221 : pp. 514-23, the entire contents of which are hereby incorporated by reference. In embodiments, one or more of the single cells from the sample are isolated and barcoded (e.g., via microfluidics, droplet encapsulation, etc.). These cells, in embodiments, undergo full transcriptome analyses (e.g., cDNA library sequencing) and / or amplicon sequencing of somatic mutations (e.g., gene capture). In embodiments, the WGS data, cDNA library sequencing, and / or gene capture sequencing is used as inputs and / or training data with one or more machine-learning processes to identify single-cell signatures.

[0052] In embodiments, the methods described herein include generating a cDNA library. In embodiments, this includes drop encapsulation of single cells, for example, where cells are counted (e.g., using microfluidic hemocytometry) and partitioned into droplets (e.g., using various droplet encapsulation techniques). In embodiments, the cells are individually sectioned within droplets, or placed into separate wells, containers, tubes, etc., and are lysed, where the mRNA is captured using unique barcodes within each individual droplet. In embodiments, the droplets are gel beads in emulsion (GEM) droplets and / or oil droplets, which can be maintained, for example, on a microfluidic system or liquid handling system adapted for single-cell usage (e.g., microfluidic chip or multi-well plate). Methods of separating and manipulating individual cells from samples are known in the art. The RNA-binding barcode, in embodiments, comprises a poly(T)-stretch (e.g., among other nucleobase patterns) which hybridizes with each mRNA at the 5’ end, 3’ end, or both ends. In some embodiments, the barcode hybridizes with each mRNA at the 3’ end, and a complementary cDNA is synthesized in the 5’ direction, recapitulating the transcript. In embodiments, the RNA-binding barcode is a DNA molecule which comprises at least a first portion which is unique to each single cell, and at least a second portion which uniquely binds and barcodes individual RNA molecules, e.g., sometimes referred to as a unique molecular identifier (UMI) which assists with accurate measurement of gene expression. This way, in embodiments, single cell gene expression patterns are trackable between individual cell datasets. In embodiments, cDNA is generated from the barcoded mRNA using reverse transcriptase (RT-PCR). The DNA barcodes, in embodiments, function as primers which allow the synthesis of cDNA, synthesized 5’-to-3’ (e.g., originating from the 3’ end of the mRNA), producing full-length DNA complements of the transcripts that faithfully recapitulate the mRNA in the cell at the time the method began. In embodiments, the mRNA and / or cDNA is additionally amplified during this process. In embodiments, the amplification occurs in a stoichiometric manner (e.g., where the ratios between cDNA abundances relate to the differences in mRNA). In embodiments, RT-PCR reagents are added to each droplet e.g., present with the barcodes in the oil / gel emulsions); these reagents include in non-limiting examples, high-fidelity reverse transcriptase and DNA polymerase, RNA / DNA nucleotides, a buffer reagent comprising magnesium ions (Mg2+), salts (e.g., NaCl), pH-buffer (e.g., Tris), and / or nuclease-free water. In embodiments, the RT-PCR droplet (or droplets, if multiple samples, or even high-throughput analyses, are being performed) is incubated using a standard thermocycler, e.g., using a RT-PCR preset for any number of cycles. In embodiments, any number of additional rounds of DNA amplification are performed to increase the cDNA yield.

[0053] In embodiments, the barcode is a DNA sequence used to distinguish between individual coding sequences present in the transcriptome. In embodiments, the DNA barcode is a randomly generated DNA barcode, which is generated by physically assembling oligos in pools. In other embodiments, the DNA barcode is a rationally designed DNA barcode, which is designed in silica, for example, using open-source design tools that are well known in the art, e.g., such as Barcode Generator (UC Davis) and nxCode (Cold Spring Harbor Laboratory), as well as commercially- available design tools (e.g., ILLUMINA, EPICENTRE, etc.). Numerous computational frameworks and design tools exist for design of DNA barcodes. It is estimated that rationally designing DNA barcodes, at 25-mer length, generates a barcode library of approximately 240,000 unique barcodes, for example, as described in Xu et al., “Design of 240,000 orthogonal 25mer DNA barcode probes.” PNAS, Vol. 106, No. 7, 2009: pp. 2289-94, the entire contents of which is hereby incorporated by reference. In embodiments, the DNA barcode is about or at least about 10 nucleotides in length (e.g., 10-mer) to about or at least about 700 nucleotides in length (e.g., 700- mer), including sizes therebetween (e.g., 25-mer, 30-mer, 35-mer, 40-mer, 50-mer, 75-mer, 100- mer, 200-mer, 500-mer, etc.). In embodiments, barcode parameters (beside length) for design considerations include the melting temperature (Tm) of the complementary portion, percent cytosine / guanidine (%G / C), the sequence complexity, and the potential cross-hybridization (e.g., percent similarity and identity stretch between probe and off-target sequences).

[0054] In embodiments, the barcoded cDNA is subjected to one or more purification steps, for example, using commercially-available DNA purification reagents (e.g., SPRISELECT reagent, AMPURE XP reagents, etc.), ethanol, buffer (e.g., buffer EB, etc.), nuclease-free water, as well as other reagents compatible with cDNA amplification and library construction. DNA purification, in embodiments, includes electrophoresis-based purification.

[0055] In embodiments, the barcoded cDNA sample is used for two or more orthogonal sequencing analyses. In embodiments, a portion of the barcoded cDNA sample is sequenced (e.g., using short-read sequencing, long-read sequencing, a combination of both, among other sequencing methods) to generate barcoded DNA sequencing data. In embodiments, another portion of the barcoded cDNA sample is used for gene capture to generate single cell gene capture sequence data. In embodiments, a first dataset is used create an initial computational model (e.g., UMAP manifold), and at least a second dataset is used to refine the model and / or annotate single cells against the model.

[0056] In embodiments, the DNA barcode further comprises a full-length P5 sequencing adaptor for the 5’ ends of the cDNA sequences in the library. In embodiments, the P5 sequencing adaptor is an oligo which functions to block the 5’ end from extraneous reaction or damage, and / or functions for complementary binding capture in a compatible flow cell (e.g., for next-generation sequencing (NGS)). In embodiments, the barcoded cDNA is further ligated to a P7 sequencing adaptor at the 3’ end of the cDNA sequences. In embodiments, the P7 sequencing adaptor is an oligo which functions to block the 3 ’ end from extraneous reaction or damage, and / or for functions for complementary binding capture in a compatible flow cell (e.g., for NGS). In embodiments using P7 adaptors, the P7 adaptor introduces a degenerate base region (DBR) which assists with identifying unique sequences, e.g., detection of duplicate sequence. P5 and P7 adaptor oligos and similar adaptor sequences are known in the art (e.g., from ILLUMINA, etc.).

[0057] In embodiments, barcoded cDNA is pooled and the cDNA library is prepped using endrepair ligation to complete the ligation of the P5:P7 adapters at the 5’ and 3’ ends, respectively. In embodiments, these adapter ends enable clear delineation of sequence reads and improve the accuracy of detecting unique, full-length sequences. In embodiments, the cDNA is processed into single-stranded cDNA and / or is fragmented and / or sheared, e.g., using vortexing / agitation, sonication, needle extrusion, restriction enzymes, etc. In some embodiments, enzymatic fragmentation is not necessary. In embodiments, single-stranded cDNA is end-prepped (e.g., fragmentation is also suitable for end-prepping). Next, in embodiments, blunt ends are ligated with a chemically modified, fully functional or half-functional (e.g., “stubby”) P7 adapter. In various embodiments, a full-length P7 is used, which comprises a poly-A sequence for construction of A- tailed cDNA libraries; alternatively, a half-functional, P7 adaptor is used. In embodiments, the P7 adapted comprises A-T overhangs for ligation. Adapter ligases for blunt-end ligation of adapters are known in the art, such as T4 DNA / RNA ligases, ssDNA ligases (e.g., CIRCLIGASE), etc. Design and use of unique P5:P7 adaptor pairs (e.g., for design of >10,000 unique adaptors) for minimizing bias and duplication detection in sequencing are well-known in the art, for example as described in Schweyen, etal., “Detection and Removal of PCR Duplicates in Population Genomic ddRAD Studies by Addition of a Degenerate Base Region (DBR) in Sequencing Adaptors,” Biol. Bull., Vol. 2272014: pp. 146-60, the entirety of which is hereby incorporated by reference.

[0058] The resulting DNA libraries, in embodiments, contain full-length (e.g., functional reads) that are end-labelled cDNA fragments (e.g., containing one or both P5:P7 adapters), and nonfunctional cDNA fragments (e.g., broken or fragmented, P7:P7 labeled). In embodiments, subsequent cDNA sequencing of the cDNA libraries enriches for P5:P7 labelled fragments, eliminating all extraneous non-functional or incomplete cDNA copies of transcripts.

[0059] In embodiments, end-pair ligation is performed with a fragmentation procedure. Fragmentation can be performed using several methods, for example in non-limiting embodiments, a controlled, site-specific enzymatic reaction using a “tagging enzyme” is employed to generate short “tags” (e.g., about or at least about 15 bp, 18 bp, 21 bp, 24 bp, or 27 bp, preferably about 21 bp in size), where these tags are ligated into cDNA sequences. In embodiments, tagging enzymes include site-specific restriction enzymes, endonucleases, or transposases, for example, Tn5 transposase which fragments double-stranded cDNA for “tagmentation” procedures. Illustrative tags include, in embodiments, P5:P7, 16S (rRNA), 18S (rRNA), and various internal transcribed spacer (ITS) sequences, for example as used in amplicon sequencing techniques (e.g., ILLUMINA and CD GENOMICS). Amplicon sequencing, in embodiments, is a highly targeted approach that enables analysis of genetic variation in specific genomic regions, where ultra-deep sequencing of PCR products (amplicons) allows efficient variant identification and characterization. Illustrative commercially-available amplicon procedures are known in the art, e.g., as produced by ILLUMINA, GENEWIZ, IDT, etc. In embodiments, adapter sequences are inserted within cDNA sequences (e.g., in addition to at the 573’ ends). In embodiments, end ligation of adapter sequences and / or internal sequence tagging are used concurrently in cDNA library sequences; alternatively, in embodiments only 573’ end adaptor ligation is used. In embodiments, the fragmentation process generates 5' and 3' “sticky ends” (e.g., overhangs) and / or blunt-ends. In embodiments, the sticky ends are converted to blunt ends, which includes removing 3' overhangs and 5'-protruded DNA ends. In embodiments, the cDNA fragments are phosphorylated at the 5' end and adenylated "A" at the 3' end to be complementary to the adapter with a thymidine "T" sticky end. In embodiments, the reaction processes include, e.g., enzymatic activities of DNA polymerases and polynucleotide kinases, for example T4 DNA polymerase, T4 polynucleotide kinase, and Taq DNA polymerase.

[0060] In embodiments, fragmentation is performed with genomic DNA to supplement the cDNA library (e.g., to supplement the transcriptome profile). Genomic DNA fragmentation, in embodiments, is performed via ligation with P5 :P7 adapters (among other forms of tagging), where the genomic DNA mixture contains full-length (e.g., functional genomic reads) that are end- labelled DNA fragments (e.g., containing both P5:P7 adapters), and non-functional genomic DNA fragments (e.g., P7:P7 labeled from broken or fragmented sequences, or sequences that were unnecessary for phenotype maintenance in the cell or tissue).

[0061] In embodiments, the methods herein provide genotype-level mutational analyses of various non-coding regions of genomic DNA that may be overlooked by transcriptional profiling alone, including for example, mutations in genes that encode for non-protein coding RNAs (e.g., tRNAs, miRNAs, mtRNAs, rRNA, etc , intronic sequences (e.g., detection of sequences spliced from transcripts, and / or identification of spliceosome issues), promoters, enhancers, silencers, insulators, structural elements of chromosomes, transcription-factor binding elements, histone- binding regions, telomeres, as well as satellite DNA. In embodiments, methods herein are suitable for identification of one or more of genomic deletions, duplications, insertions, or indels, including translocations, frameshifts, stop codon insertion, and nucleotide variants which result in protein mutations. Various genomic DNA elements that contribute to disease which are amendable to mutational analyses described herein, are known in the art, for example as described in the ENCODE Project Consortium, “An integrated encyclopedia of DNA elements in the human genome,” Nature, Vol. 489, No. 7414, 2012: pp. 57-74; and in Maston et al., “Transcriptional regulatory elements in the human genome,” Annu Rev Genomics Hum Genet, Vol. 7, 2006: pp. 29- 59, the entirety of each of which is hereby incorporated by reference.

[0062] In embodiments, methods herein utilize DNA hybridization capture, also referred to herein as “gene capture,” or “hybrid gene capture.” In embodiments, gene capture methods utilize the barcoded cDNA along with additional DNA probes, e.g., such as commercially-available custom hybridization capture panels from TWIST BIOSCIENCES, ILLUMINA, INTEGRATED DNA TECHNOLOGIES (IDT), AGILENT, etc. Hybridization capture panels are known in the art {e.g., TWIST HUMAN CORE EXOME KIT), and in various embodiments, are customizable to cover mutation-specific genes, disease-specific genes, hematopoietic-specific panels, stem cell panels, tissue-specific panels, etc.

[0063] In embodiments, the gene capture probes are single-stranded or double-stranded DNA probes. In embodiments, the capture probes are utilized with any fragmentation procedure, as described herein. In embodiments, capture probes are amenable to design tools and strategies congruent with the design of barcodes, as described herein. In embodiments, gene capture probe specificity and sensitivity is tunable by modifying probe parameters, including for example, the probe length {e.g., about 25-mer, about 30-mer, about 35-mer, about 40-mer, about 50-mer, about 75-mer, about 100-mer, etc.), the melting temperature (Tm) of the complementary portion, percent cytosine / guanidine (%G / C), and the sequence complexity, and evaluating the potential crosshybridization {e.g., percent similarity and identity stretch between probe and off-target sequences).

[0064] In embodiments, the barcoded cDNA has one or more adaptor or tag, e.g., at the 5’ end, and is then bound by a complementary sequence in the capture probe at the 3’ end. In embodiments, the capture probe is chemically -modified with biotin or streptavidin {e.g., biotinylated at its 5’ end) and is thus suitable to being captured via a capture media containing the cognate biotin or streptavidin. In embodiments, the capture probes are biotinylated, and the capture media is streptavidin-coated. In embodiments, the capture media is a bead, plate, well, tube, or microarray chip. In embodiments, each cDNA molecule is captured via a biotin-streptavidin linkage. In embodiments, the capture media is a magnetic bead that is manipulatable by application of a magnetic field {e.g., with a magnet). Methods of gene capture herein are compatible with cDNA generated from mRNA and / or genomic DNA without any particular prior barcoding. In embodiments, probe-bound cDNA is captured using a magnet, washed, and target genes are then PCR amplified. In embodiments, the resultant enriched, amplified cDNA undergoes adaptor ligation, e.g., as described herein, for long read sequencing (e.g., using a PACBIO sequencing platform). In embodiments, the gene capture sequence data is processed such that mutation information from each read is linked to the original cell barcode (e.g., 10XGENOMICS barcodes). In embodiments, this generates a list of cell barcodes with mutation information that is used to annotate the UMAP manifold that was initially generated.

[0065] In embodiments, DNA, cDNA libraries, and / or gene capture sequences are amplified using various PCR methods. In embodiments, this includes nested PCR methods, which is useful to increase the sensitivity and / or specificity, for example, when amplifying a particular member of a polymorphic gene family or when amplifying a cDNA copy of an mRNA present at very low abundance in a clinical specimen containing a heterogeneous population of cell types. In embodiments, nested PCR involves two sequential amplification reactions, each of which uses a different pair of primers. In embodiments, the product of a first amplification reaction is used as the template for a second round of PCR, which is primed by oligonucleotides that are placed internal to the first primer pair. The use of two pairs of oligonucleotides allows a higher number of cycles to be performed, thereby increasing the sensitivity of the PCR. In embodiments, the improved specificity of the reaction derives from the binding of two separate sets of primers to each target template. In embodiments, nested PCR is an efficient method to amplify segments of long templates but requires some knowledge of the sequence of the target.

[0066] In embodiments, the methods described herein further comprise sequencing the barcoded cDNA library and / or gene capture cDNA (e.g., using index PCR). In embodiments, the methods described herein include the use of one or more commercially-available DNA sequencers known in the art, including in non-limiting examples, one or more of ILLUMINA (e.g., NOVASEQ X, MISEQ, HISEQ), PACBIO (e.g., SEQUEL I and II systems), or OXFORD NANOPORE TECHNOLOGIES (e.g. , MINION), with exemplary methods as described, for example, in Hafner, et. al., “Barcoded cDNA library preparation for small RNA profiling by next-generation sequencing,” Methods, Vol. 58, No. 2, 2012: pp. 164-70; Mikheenko, et al., “Sequencing of individual barcoded cDNAs using Pacific Biosciences and Oxford Nanopore Technologies reveals platform-specific error patterns,” Genome Res., Vol. 32, 2022: 726-37; and Wang, et al., “Nanopore sequencing technology, bioinformatics, and applications,” Nature Biotechnology, Vol. 39, 2021 : pp. 1348-65, each of which is hereby incorporated by reference in their entirety. Persons skilled in the art, with the benefit of this filing in its entirety, will be aware of the various DNA / RNA sequencers and sequencing methods known in the art suitable for methods herein.

[0067] In embodiments, the methods described herein utilize sequencing data generated from short-read and / or long-read sequencing. In embodiments, the methods described herein use a combination of hybrid gene capture data and long read sequencing data which captures the distal mutations of the cell, and allows for custom probe design, multiplex targeting, and reduced PCR error. In embodiments, single cell genotyping methods utilize nested PCR and short read sequencing, which allow for high specificity and sensitivity, high coverage, and flexible and inexpensive sequencing. Without wishing to be bound by theory, in embodiments, long read hybrid gene capture, and its various combinations in use, determine single cell genotype with greater accuracy than nested PCR alone.

[0068] In embodiments of the methods herein, sequencing data relates to nucleic acids consisting of DNA and RNA. In embodiments, the RNA is selected from the coding and non-coding RNA. In embodiments, single cell expression datasets are generated using one or more method of singlecell RNA sequencing (scRNA-seq), scTag-seq, single-cell assay for transposase-accessible chromatin using sequencing (scATAC-seq), CyTOF / SCoP, E-MS / Abseq, miRNA-seq, CITE-seq and any combinations or summaries thereof.

[0069] The data, in embodiments, contains a series of nucleic acid sequences (e.g., DNA) which relate to the gene expression (e.g., mRNA) present in the cell at the time the method began. In embodiments, this data comprises one or more of sequences of nucleic acid bases, identifying information relating to a gene identification, a wild-type paired sequence (e.g, including homology, similarity, or other characteristics), a barcode sequence or unique identifier, and / or a cell-specific identifier sequence. In embodiments, with respect to machine-learning, “model,” “map,” “projection,” and “manifold,” are used interchangeably to refer to one or more outputs of the machine-learning algorithm / process.

[0070] In embodiments, methods herein utilize sequencing data with one or more computer-based machine-learning algorithms / methods, such as unsupervised or supervised clustering machine learning processes. These methods, in embodiments, are typically nonlinear, dimension reduction techniques that are useful for calculating local, low-dimensional (e.g., 2-dimensional or 3- dimensional) representations of sequencing data, particularly useful for visualizing how individual data points are related to one another (e.g., as shown in Figs. 2A - 2C and 4). Despite their appearance, these clustering outputs, in embodiments, are not Gaussian, meaning their parameters (e.g., distances between points) are relative and lack defined numerical relationships in the visual representation.

[0071] In embodiments, the clustering machine-learning process is selected from uniform manifold approximation and projection (UMAP), t-distributed stochastic neighbor embedding (t- SNE), and principal component analysis (PCA). Persons skilled in the art, with the benefit of this disclosure in its entirety, will be aware of the various machine-learning processes that are suitable for clustering of data and the analyses as described herein.

[0072] In embodiments, illustrative open-source machine-learning algorithms for methods herein include, in non-limiting examples, “umap-leam” script package for UMAP analysis, “FItSNE” script package for t-SNE analysis, and “scvis” script package for PCA, each of which is compatible with py-based programming (e.g., pythons'). The theoretical mathematical framework of UMAP algorithms, including simplified algorithm outlines for programming purposes, are well-known in the art, for example, as described in Mclnnes, et al. (2018) “UMAP: Uniform Manifold Approximation and Proj ection for Dimension Reduction,” ArXiv e-prints, 1802.03426, the entirety of which is hereby incorporated by reference. In embodiments, methods herein incorporate UMAP for single cell analyses performed substantially as described in Becht, et al., “Dimensionality reduction for visualizing single-cell data using UMAP,” Nature Biotechnology, Vol. 37, No. 1, 2019: pp. 38-45. In embodiments, methods herein incorporate UMAP analyses with densMAP augmentation which better preserves local density information (e.g., through the addition of k- nearest neighbors), which ultimately improves visualization of heterogeneity present in transcriptomic variability, for example, performed as substantially described in Narayan, et al., “Assessing single-cell transcriptomic variability through density-preserving data visualization,” Nature Biotechnology, Vol. 39, No. 6, 2021: pp. 765-74, the entire contents of which are hereby incorporated by reference. In embodiments, a combination of UMAP and densMAP provides better spread of densely packed sequencing datapoints which improves visualization and identification of mutations and / or polymorphisms. In embodiments, UMAP analysis which is an algorithm for dimension reduction based on manifold learning techniques and ideas from topological data analysis is used. In embodiments, this visualization not only provides a general framework for approaching manifold learning and dimension reduction, but also provides specific concrete realizations. In embodiments, t-SNE analysis which is another dimensionality reduction technique (unsupervised machine-learning algorithm), able to use sequencing data inputs that lack labels is used. In embodiments, PCA which is an unsupervised machine-learning algorithm typically used in a variety of computational biology applications, is used.

[0073] Dimensionality reduction, in embodiments, is performed on datasets to generate dimension reduction components (e.g., in the form of matrix M). In embodiments, performing dimensionality reduction includes performing principal component analysis (PCA) on the single cell expression datasets to generate dimension reduction components. In embodiments, performing dimensionality reduction is accomplished using diffusion maps and / or neural network autoencoders on the datasets to generate dimension reduction components.

[0074] Similarly, in embodiments, manifold learning is performed using one or more dimension reduction components (e.g., in the form of matrix AT) to create another data form, such as the matrix N. In embodiments, performing manifold learning includes estimating a geometry of the data in matrix AT to create the matrix N. In such embodiments, performing manifold learning can include performing locally linear embedding, local linear isometric mapping (ISOMAP), t-SNE, Potential of Heat-Diffusion for Affinity Based Trajectory Embedding (PHATE), or UMAP. In embodiments, performing manifold learning includes creating a force-directed layout based on the data in matrix AT to generate the matrix N. In embodiments, a force-directed layout is created using the Force Atlas 2 algorithm.

[0075] In embodiments, performing clustering assumes no prior knowledge of the organization of the plurality of points in each cluster. In embodiments, clustering includes performing HDBSCAN to generate the set of clusters Cj, for example as described in Campello, et al., (2015), “Hierarchical density estimates for data clustering, visualization, and outlier detection,” ACM Transactions on Knowledge Discovery from Data (TKDD), Vol. 10, No. 1, the entire contents of which are hereby incorporated by reference. In embodiments, clustering includes performing Louvain community detection to generate the set of clusters Cj, for example as described in Blondel, et al. (2008), “Fast unfolding of communities in large networks,” Journal of statistical mechanics: theory and experiment, 2008(10), Pl 0008, the entire contents of which are hereby incorporated by reference. In embodiments, clustering includes the performance of both HDBSCAN and Louvain community detection.

[0076] In embodiments, performing clustering includes assigning each point to one of the clusters Cj based on a point in time at which the single cell expression dataset associated with the point was collected. In embodiments, performing clustering includes analyzing the plurality of points using a diffusion path algorithm that assigns a point to a cluster based on a measure of how terminal the point is from the cluster.

[0077] In embodiments, methods herein include performing dimensionality reduction on a first dataset and / or a second dataset, where each comprises single cell expression data. This dimensionality reduction, in embodiments, generates one or more dimension reduction components. In embodiments, the dimension reduction comprises one or more of a uniform manifold approximation and projection (UMAP) algorithm, a principal component analysis (PCA) algorithm, a random projection algorithm, an independent component analysis algorithm, a feature selection method, a factor analysis algorithm, Sammon mapping, curvilinear components analysis, a stochastic neighbor embedding (SNE) algorithm, an local linear isometric mapping (ISOMAP) algorithm, a maximum variance unfolding algorithm, a locally linear embedding algorithm, a t- distributed stochastic neighbor embedding (t-STSTE) algorithm, a non-negative matrix factorization algorithm, a kernel principal component analysis algorithm, a graph-based kernel principal component analysis algorithm, a linear discriminant analysis algorithm, a generalized discriminant analysis algorithm, a Large Vis algorithm, a Laplacian Eigenmap algorithm, or a Fisher's linear discriminant analysis algorithm, for example, as described in Fodor, “A survey of dimension reduction techniques,” Center for Applied Scientific Computing, Lawrence Livermore National, Technical Report, (2002), UCRL-ID- 148494; Cunningham, “Dimension Reduction,” University College Dublin, Technical Report, (2007), UCD-CSL2007-7, Zahorian and Hu, “Nonlinear Dimensionality Reduction Methods for Use with Automatic Speech Recognition,” Speech Technologies, (2011), ISBN 978-953-307-996-7; and Padmaja and Vishnuvardhan, “2016 IEEE 6th International Conference on Advanced Computing (IACC),” 2016: pp. 31-34, ISBN 978-1- 4673-8286-1, each of which is hereby incorporated by reference in their entirety. Accordingly, in embodiments, the dimensionality reduction includes PCA, and each respective extracted dimension reduction component comprises a respective principal component derived by the PCA. In such embodiments, the number of principal components in the plurality of principal components is limited to a threshold number of principal components calculated by the PCA algorithm. In embodiments, the threshold number of principal components comprises, for example, 5, 10, 20, 50, 100, 1000, 1500, or any other number. In embodiments, each principal component calculated by the PCA algorithm is assigned an eigenvalue by the PCA algorithm, and the corresponding subset of the first plurality of extracted features is limited to the threshold number of principal components assigned the highest eigenvalues. In embodiments, for each respective cellular-component vector, the plurality of dimension reduction components are applied to the respective cellular-component vector to form a corresponding dimension reduction vector that includes a dimension reduction component value for each respective dimension reduction component in the plurality of dimension reduction components. In embodiments, this forms a corresponding first and second plurality of dimension reduction vectors. In embodiments, the method includes performing clustering to generate a set of clusters Cj, where each cluster includes a plurality of points corresponding to a subset of the first and second plurality of dimension reduction vectors. In embodiments, the first plurality of cells from a first cluster of the set of clusters Cj and the second plurality of cells from a second cluster of the set of clusters Cj are both identifiable.

[0078] In embodiments, methods herein include performing manifold learning with a corresponding first and second plurality of dimension reduction vectors. In embodiments, this manifold learning identifies a relative cell state of each cell with respect to each other cell in the first and second plurality of cells, for example as described in Wang and Zha, “Adaptive Manifold Learning,” Advances in Neural Information Processing Systems, Vol. 17, 2004: pp. 1-8, which is hereby incorporated by reference in its entirety.

[0079] In embodiments, methods herein include performing a UMAP machine-learning process, using a first dataset of sequencing data, and then refining the projection with a second dataset. In embodiments, the UMAP projection is generated with the sequencing data of the cDNA library sequencing (e.g., using the P5:P7 adapter ligated cDNA set). In embodiments, this provides a data visualization (e.g., such as those in Figs. 2A-2C and 4) which clusters datapoints (e.g., individual sequences) based on a variety of similarity factors. The refinement process, in embodiments, adds additional dimensionality, for example, where a z-axis emerges based on the relative mRNA level. In embodiments, such dimensionality provides gene expression information such as the overexpression of wild-type sequences, in additional to identifying mutations or deviations from a wild-type sequence. In embodiments, mutants or other disease-relevant sequence information is identifiable by comparison of two or more UMAP projections of single cells, where the cells differ in one or more features (e.g., cancer cell vs. healthy cell / wild-type, cell prior to treatment vs. after treatment, etc.).

[0080] In embodiments, refinement of outputs (e.g. analysis of curated and demultiplexed sequencing data), include but are not limited to, cluster annotation, correlation activity analysis, leave-one-out classifier analysis, nearest-neighbor analysis (e.g., densMAP), and differential expression analysis. For example, in embodiments, cluster annotation includes annotating curated datasets using defined markers to score cells and assign cells to a specific cell cluster present in the sequenced cell population. Non-limiting methods for performing cluster annotation include PCA, and non-limiting examples of methods useful for visualizing the clusters include UMAP. In embodiments, after cluster annotation is performed, the resulting data is aggregated (e.g., pseudobulk data) and / or v-scores are computed.

[0081] Within single cell datasets, in embodiments, these factors include clustering sequencing datapoints based on gene function (e.g., metabolic genes, ubiquitin-proteasome genes, cell cycle genes, endocytic vesicular system genes, etc.), based on subcellular localization (e.g., cell surface marker genes, organelle-specific genes, etc.), based on draggability or target attractiveness (e.g., G-coupled receptor proteins, kinases, etc ), based on cell-pathway specific genes (e.g., JAK / STAT, ATR, ATM, ERK, hormone-signaling, etc.), etc. In embodiments, a mutation in a gene of interest is identifiable based on its location in the UMAP projection, and / or the function of a gene of interest is identifiable based on its location in the UMAP projection.

[0082] Where heterogeneous cell samples are concerned, in embodiments, categorization is based on cell-specific gene expression patterns, for example, a dataset for immune cells would comprise datapoints clustered according to the identified sequences as being categorizable based on belonging to individual cell types, such as neutrophils, plasma cells, B cells, T cells, dendritic cells, platelets, megakaryocytes, mast cells, monocytes, basophils, myeloid precursor cells, etc. In embodiments, given a sufficient number of datapoints, subsets of each cluster appear (e.g., especially in a UMAP projection with densMAP to spread out datapoints), for example, where among the T cell categorization, individual T cell subsets emerge, e.g., CD4+ T cell, CD8+ T cell, CD7+ T cell, aP T cell, y5 T cell, naive T cell, etc. In embodiments, differences among individual cells are identifiable based on the clustering patterns of the UMAP projection, and / or based on the locations of individual datapoints / clusters in the UMAP projection.

[0083] In embodiments, annotating the single cell genotype comprises classifying the single cell into one or more categories based on the identification of one or more mutations. In embodiments, the one or more categories comprises a healthy (e.g., a wild-type) category and a diseased (e.g., abnormal) category. The one or more categories, in embodiments, are informed by gene expression and / or one or more additional cellular factors (e.g., cell type, differentiation potential, cell surface protein expression, biomarkers, etc.). In embodiments, the one or more categories relate to one or more chemi cal -genetic relationships, for example, for selection of a treatment intended to ameliorate a cellular phenotype driven by one or more identified mutations in the cell.

[0084] In embodiments, the quality of the UMAP analysis (e.g., the clustering and spread of datapoints) changes as a function of the dataset that is input and / or the dataset that is used to refine and / or annotate cells. For example, in reference to Fig. 2A - 2C, the clustering observed in a projection generated from nested PCR Seq versus long read sequencing may alter datapoint positioning and local density of datapoints. In embodiments, two or more different biochemical techniques are used to obtain sequence information on single cells to result in more robust UMAP projections.

[0085] In embodiments, the methods of the present disclosure provide improved detection of a gene signature, e.g., as compared to detection of a gene signature using single cell genotyping alone. In embodiments, the methods of the present disclosure provides improved detection of a gene signature, e.g., as compared to detection of a gene signature using single cell gene expression alone.

[0086] In embodiments, the methods of the present disclosure provide improved detection of cell phenotyping, e.g., as compared to detection of a gene signature using single cell genotyping alone. In embodiments, the methods of the present disclosure provide improved detection of cell behavior (e.g., functionality, metabolism, signaling, homeostasis, stimuli responsiveness, etc.), e.g., as compared to detection of a gene signature using single cell gene expression alone. In embodiments, the methods of the present disclosure comprise single cell genotyping and single cell gene expression for improved detection of a gene signature and prediction of cell mutations, e.g., as compared to using single cell genotyping alone. In embodiments, the methods of the present disclosure comprise single cell genotyping and single cell gene expression for improved detection of a gene signature and predictions of cell mutations, e.g., as compared to using single cell gene expression alone.

[0087] In embodiments, the single cell genotyping approach described herein is demonstratable in a mixed sample where genotypes are correctly assigned to known cells, with genotype well correlated to expression pattern. For example, in non-limiting embodiments, a cell line derived from a subject is used, e.g., where the subject is diagnosed with, or suspected of a diagnosis, of a hematological proliferative disorder harboring a driver mutation in JAK2. In such non-limiting embodiments, methods herein comprise using a matched cell line that is wild-type at the JAK2 locus. In embodiments, mutation calls are confirmed in parallel using a published nested PCR method (e.g., AMPLISEQ from ILLUMINA); all confident mutation calls agreed between the two methods across overlapping cell barcodes. In embodiments, a proportion of genotypes is found to be concordant between the hybridization-based method and the amplification-based method, and a plurality of genes, including without limitation, target genes and one or more driver genes (e.g., “driving” cell phenotype), from individual cells are simultaneously profiled.

[0088] In embodiments, method of the present disclosure comprises analyzing, in parallel, multiple genes with similar transcriptional levels, which additionally provides method scalability. In embodiments, methods of the present disclosure do not require specialized probes for identification of DNA or RNA. In embodiments, methods of the present disclosure provide cost- effective long read sequencing approaches compared to short read sequencing. In embodiments, methods of present disclosure provide for improved de novo assembly, mapping certainty, transcript isoform identification, and / or detection of structural variants.

[0089] Methods of Measuring Disease Progression and Informing Treatment

[0090] In aspects, the disclosure provides methods for measuring disease progression by analyzing refined cell signatures. In embodiments, cell signatures are refined using the single cell transcriptome and genotype information generated by the methods described herein. In embodiments, the disclosure provides methods for determining disease signature and / or stratifying a subject based on a specific mutation-carrying cell or mutation identification. In embodiments, methods herein are useful for measuring disease progression by annotating one or more single cell genotypes using one or more models developed as described herein.

[0091] In embodiments, the methods described herein include obtaining a cell or clonal cell population, e.g., via a blood draw or biopsy, for example, where the blood draw comprises circulating tumor cells or the biopsy is a tumor biopsy. Persons skilled in the art will appreciate methods herein are suitable for any cell, tissue, or sample obtained from an individual, such as blood, plasma, lymph, cerebrospinal fluid, interstitial fluid, etc.

[0092] In embodiments, the methods described herein provide time-dependent transcriptomic variability which relates to disease phenotype, which ultimately translate to subject stratification and / or treatment selection. In embodiments, methods herein utilize cell expression profiles and genotyping which provide data that relates to temporal differences in cell expression and mutation accumulation, this provides, in embodiments, mutational kinetics, which relate to relative velocities of mutational change, e.g., as it relates to particular genes, genomic “hotspots,” or cell signaling pathways.

[0093] In embodiments, one or more cells are obtained at a first time point and again at a later time point, where one or more differences in cell signature (e.g., gene expression, genotype, accumulation of somatic mutants, protein expression, mRNA levels, etc.) are identified between time points. These time points, in embodiments, differ in disease state (e.g., a malignant sample vs. a sample taken once a subject enters remission; a sample taken in an early stage of a disease vs. a later stage of a disease), or differ in treatment state (e.g., a sample obtained before treatment begins vs. at some point after treatment starts). In embodiments, temporal differences in cell expression profiles and genotyping provide information about the development of drug resistance, the development of refractory disease, and in identifying the establishment of remission.

[0094] In embodiments, a time between timepoints comprises about or at least about 1 day, about or at least about 2 days, about or at least about 3 days, about or at least about 1 week, about or at least about 2 weeks, about or at least about 3 weeks, about or at least about 1 month, about or at least about 2 months, about or at least about 6 months, about or at least about 1 year, about or at least about 2 years, or about or at least about 5 years, including time points thereafter and therebetween. In embodiments, at least a first single cell genotype is annotated at a first timepoint, and at least a second single cell genotype is annotated at a second timepoint, where a difference in one or more mutations (e.g., substitutions, variants, deletions, duplications, insertions, translocations, or indels), and / or single nucleotide polymorphisms between the first timepoint and second timepoint indicates a progression in disease state. In embodiments, the difference is an accumulation of one or more mutations between timepoints. Alternatively, in embodiments, the difference is the absence of one or more mutations between timepoints (e.g., loss of a druggable target). In some embodiments, the difference is both the accumulation and loss of one or more mutations between timepoints.

[0095] In embodiments, the methods described herein use mutation information identified in cell signatures as an input for identifying one or more relevant therapeutic agents targeting mutationspecific cell populations. In embodiments, these chemi cal -genetic approaches are based on using the cell signatures (e.g., gene expression and genotyping) as a input into one or more processes which index therapeutic agents (e.g., various small molecules, etc ) based on purported chemical- genetic effects.

[0096] For example, in embodiments, the present methods include performing correlation activity analysis, which includes determining whether a compound induces a transcriptional signature (e.g., compound-specific differential transcriptional signature), induces a difference from a control sample (e.g., DMSO) in a statistically significant manner. In embodiments, the difference is a difference in cell phenotype, including differences in cell viability, gene expression, mRNA level, or protein expression. The correlation activity analysis, in embodiments, includes the use of one or more statistical models. In non-limiting embodiments, a correlation between the compoundspecific differential transcriptional signature and the transcriptional signature of the control sample is computed and / or a pairwise correlation between the transcriptional signature of each control sample is computed. In embodiments, this includes statistical analyses, including one or more of a difference of means test, a Wilcoxon rank-sum test, a t-test, a logistic regression, or a generalized linear model, performed on the data resulting from each correlation. In embodiments, v-scores are generated, where the v-scores are utilized in predictive model to determine whether the model correctly classifies and identifies the model. Non-limiting examples of predictive models are found in U.S. Patent Publication No. US 2020 / 0020419, which is incorporated by reference herein in its entirety. In embodiments, the methods described include performing a leave-one-out classifier analysis, which includes distinguishing a compound-specific differential transcriptional signature from the compound-specific differential transcriptional signatures of one or more compounds that were experimentally determined at the same time (e.g. each of the compounds were included in the same plate). In non-limiting embodiments, compound-specific differential transcriptional signatures for each compound are determined in replicate (e.g., two, three, four, or four or more replicates per compound) with a discrete label assigned to the compound-specific differential transcriptional signatures obtained from each replicate, and the consistency and / or reproducibility of the obtained transcriptional signatures is assessed. In embodiments, methods for assessing the consistency and / or reproducibility of the obtained compound-specific differential transcriptional signature include, but are not limited to, training a logistic regression classifier on the compoundspecific differential transcriptional signature obtained in one of the replicates and using the classifier to classify the compound-specific differential transcriptional signature of the remaining replicates.

[0097] In embodiments, the present methods include performing a nearest-neighbor analysis, which includes assessing the consistency and / or reproducibility of compound-specific differential transcriptional signatures by processing v-scores of the compound-specific differential transcriptional signature for non-control samples of one or more compounds, where the transcriptional signatures are experimentally determined orthogonally e.g., each of the compounds were included within the same screen). In non-limiting embodiments, methods to assess the consistency and / or reproducibility of compound-specific differential transcriptional signatures include generating a confusion plot.

[0098] In embodiments, the present methods include performing a differential expression analysis, which includes determining which genes are differentially regulated between one or more compounds and control samples. In non-limiting embodiments, methods for performing differential expression analysis include DEseq2.

[0099] In embodiments, the present methods herein include stratifying a subject as a function of the progression in disease state, for example, where stratifying comprises classifying the subject into a diagnosis category. For example, in embodiments, diagnoses categories include, but are not limited to, solid tumor, hematological cancers, acquired diseases, congenital diseases, hematopoietic and non-hematopoietic diseases, and diseases affecting neurons, astrocytes, oligodendrocytes, cardiomyocytes, skeletal muscle cells, hepatocytes, pancreatic P cells, lung epithelial cells, etc. In embodiments, stratification of subjects is based on mutation categorization, for example, where one or more mutants identified in one or more cells originating from the subject are related to their cognate signaling pathway and its potential role in disease, as has been done in non-limiting examples, with BCR-ABL1 fusion protein translocation in leukemias (CML), BRAC1 / 2 gene mutants in breast cancers, endometrial cancers, melanomas, pancreatic and prostate cancer, and JAK2 mutants (e.g., V617F) in HSC populations, etc.

[0100] In embodiments, stratification of subjects and / or driving mutation is based on cancer type, for instance in non-limiting examples, to a category selected from carcinoma, sarcoma, myeloma, leukemia, lymphoma, mixed-type cancer, and / or metastatic cancer. In non-limiting embodiments, the stratification and / or classification of one ore more mutations or subjects is selected from one or more categories of Acute Biphenotypic Leukemia, Acute Eosinophilic Leukemia, Acute Lymphoblastic Leukemia, Acute Lymphocytic Leukemia, Acute Myeloid Dendritic Cell Leukemia, Acute Myeloid Leukemia, Adenocarcinoma of the Lung, Adrenal Gland Tumors, Adrenocortical Carcinoma, AIDS-related Cancers, AIDS-related Lymphoma, Alveolar Soft Part and Cardiac Sarcoma, Amyloidosis, Anal Cancer, Anaplastic Large Cell Lymphoma, Angioimmunoblastic T-cell Lymphoma, Appendix Cancer, Astrocytoma, Ataxia-telangiectasia, Attenuated Familial Adenomatous Polyposis, B-cell Prolymphocytic Leukemia, Basal Cell Carcinoma, Beckwith-Wiedemann Syndrome, Bile Duct Cancer, Birt-Hogg-Dube Syndrome, Bladder Cancer, Bone Cancer, Brain and Nervous System Cancer, Brain Stem Glioma, Brainstem Glioma, Brain Tumors, Breast Cancer, Bronchial Adenomas / Carcinoids, Burkitt's Lymphoma, Carcinoid Tumor, Carcinoid Tumors, Carney Complex, Central Nervous System Tumors, Cerebellar Astrocytoma, Cerebral Astrocytoma, Cervical Cancer, Childhood Desmoplastic Ganglioglioma, Cholangiocarcinoma, Chondrosarcoma, Chronic Lymphocytic Leukemia, Chronic Myelogenous Leukemia, Chronic Myeloid Leukemia, Chronic T-cell Lymphocytic Leukemia, Colon Cancer, Colorectal Cancer, Cowden Syndrome, Craniopharyngioma, Cutaneous T-cell Lymphoma, Dermatofibrosarcoma Protuberans, Desmoplastic Small Round Cell Tumor, Diffuse Gastric Cancer, Diffuse Large B-cell Lymphoma, Endocrine System Cancer, Endocrine Tumors, Endometrial Cancer, Eosinophilic Leukemia, Ependymoma, Epithelioid Hemangioendothelioma (EHE), Esophageal Cancer, Ewing Sarcoma, Extragonadal Germ Cell Tumor, Extrahepatic Bile Duct Cancer, Eye Cancer, Eyelid Cancer, Fallopian Tube Cancer, Familial Adenomatous Polyposis, Familial Malignant Melanoma, Familial Clear Cell Renal Cell Carcinoma (RCC), Follicular Lymphoma, Gallbladder Cancer, Gardner Syndrome, Gastric (stomach) Cancer, Gastrointestinal Cancer, Gastrointestinal Carcinoid Tumor, Gastrointestinal Stromal Tumor, Gastrointestinal Stromal Tumor (GIST), Genitourinary and Gynecologic Cancer, Germ Cell Tumor, Gestational Trophoblastic Disease, Gestational Trophoblastic Tumor, Glioblastoma, Glioma, Hairy Cell Leukemia, Head and Neck Cancer, Hematopoietic Cancer, Hepatocellular Cancer, Hepatosplenic T-cell Lymphoma, HIV-related Cancer, Hodgkin's Lymphoma, Hypopharyngeal Cancer, Inflammatory Breast Cancer, Intravascular Large B-cell Lymphoma, Invasive Cribriform Carcinoma, Invasive Lobular Carcinoma, Islet Cell Carcinoma (endocrine Pancreas), Islet Cell Tumors, luvenile Polyposis Syndrome, Kaposi Sarcoma, Keratoacanthoma, Kidney Cancer, Lacrimal Gland Tumor, Large Granular Lymphocytic Leukemia, Laryngeal and Hypopharyngeal Cancer, Leiomyomatosis and Renal Cell Cancer, Leiomyosarcoma, Li-Fraumeni Syndrome (LFS), Liposarcoma, Liver Cancer, Lung Cancer, Lymphomas of Primary Cutaneous Origin, Lymphomatoid Granulomatosis, Lymphoplasmacytic Lymphoma, Lynch Syndrome, Malignant Fibrous Histiocytoma of Bone, Mantle Cell Lymphoma, Marginal Zone B-cell Lymphoma, Mast Cell Leukemia, Mastocytosis, Mediastinal Large B Cell Lymphoma, Medullary Carcinoma, Medulloblastoma, Melanoma, Meningioma, Merkel Cell Carcinoma, Mesothelioma, Mixed Polyposis Syndrome, Mucosa-associated Lymphoid Tissue Lymphoma, Muir-Torre Syndrome (MTS), Multiple Endocrine Neoplasia Syndrome, Multiple Endocrine Neoplasia Type 1, Multiple Endocrine Neoplasia Type 2, Multiple Myeloma, Multiple Myeloma / Plasma Cell Neoplasm, Mycosis Fungoides, Myelodysplastic Syndromes, MYH- associated Polyposis, Myxosarcoma, Nasal Cavity and Paranasal Sinus Cancer, Nephroblastoma, Nasopharyngeal Cancer, Nasopharyngeal Carcinoma, Neuroblastoma, Neuroendocrine Tumors, Neurofibromatosis Type 1, Neurofibromatosis Type 2, Nevoid Basal Cell Carcinoma Syndrome, Nodal Marginal Zone B Cell Lymphoma, Non-Hodgkin Lymphoma, Non-small Cell Lung Cancer, Non-small Cell Lung Carcinoma, Oligodendroglioma, Optic Nerve Glioma, Oral and / or Oropharyngeal Cancer, Osteosarcoma, Ovarian Cancer, Ovarian Epithelial Cancer, Ovarian Germ Cell Tumor, Pancreatic Cancer, Papillary Renal Cell Carcinoma, Paranasal Sinus and Nasal Cavity Cancer, Parathyroid Cancer, Pelvic Cancer, Penile Cancer, Peutz-Jeghers Syndrome, Pharyngeal Cancer, Pheochromocytoma, Phyllodes Tumor, Pilocytic Astrocytoma, Pineal Astrocytoma, Pituitary Adenoma, Pituitary Gland Tumors, Plasmablastic Lymphoma, Pleuropulmonary Blastoma, Precursor B Lymphoblastic Leukemia, Primary Central Nervous System Lymphoma, Primary Cutaneous Follicular Lymphoma, Primary Cutaneous Immunocytoma, Primary Effusion Lymphoma, Primitive Neuroectodermal Tumor, Prostate Cancer, Rectal Cancer, Renal Cancer, Renal Cell Carcinoma, Retinoblastoma, Rhabdomyosarcoma, Salivary Gland Cancer, Sarcoma, Sarcomas of Primary Cutaneous Origin, Sebaceous Carcinoma, Sezary Syndrome, Skin Adnexal Tumors, Skin Cancer, Small Bowel Cancer, Small Cell Lung Cancer, Small Intestine Cancer, Splenic Marginal Zone Lymphoma, Squamous-cell Carcinoma of the Lung, Squamous Cell Carcinoma, Squamous Cell Skin Cancer, Stomach Cancer, Surface Epithelial-Stromal Tumor, T- cell Prolymphocytic Leukemia, Testicular Cancer, Thoracic and Respiratory Cancer, Thymoma, Thymic Carcinoma, Thyroid Cancer, Transitional Cell Cancer, Transitional Cell Canceradenoid Cystic Carcinoma, Tuberous Sclerosis Syndrome, Tubular Carcinoma, Turcot Syndrome, Unknown Primary Cancer, Unsorted Cancer, Ureter Cancer, Urethral Cancer, Uterine Cancer, Uterine Sarcoma, Uveal Melanoma, Vaginal Cancer, Visual Pathway and Hypothalamic Glioma, Von Hippel-Lindau (VHL) Syndrome, Vulvar Cancer, Wilms Tumor, and Xeroderma Pigmentosum.

[0101] In non-limiting embodiments, the stratification and / or classification of one or more mutations or subjects is selected from one or more categories associated with adenoid cystic carcinoma, adrenal gland tumors, amyloidosis, anal cancer, appendix cancer, astrocytoma, ataxiatelangiectasia, attenuated familial adenomatous polyposis, Beckwith-Wiedemann Syndrome, bile duct cancer, Birt-Hogg-Dube Syndrome, bladder cancer, bone cancer, brain stem glioma, brain tumors, breast cancer, carcinoid tumors, Camey complex, central nervous system tumors, cervical cancer, colorectal cancer, Cowden syndrome, craniopharyngioma, desmoplastic infantile ganglioglioma, endocrine tumors, ependymoma, esophageal cancer, Ewing sarcoma, eye cancer, eyelid cancer, fallopian tube cancer, familial adenomatous polyposis, familial malignant melanoma, familial non-VHL clear cell renal cell carcinoma, gallbladder cancer, Gardner Syndrome, gastrointestinal stromal tumor, germ cell tumor, gestational trophoblastic disease, head and neck cancer, diffuse gastric cancer, leiomyomatosis and renal cell cancer, mixed polyposis syndrome, pancreatitis, papillary renal cell carcinoma, islet cell tumors, juvenile polyposis syndrome, kidney cancer, lacrimal gland tumor, laryngeal and hypopharyngeal cancer, acute lymphoblastic leukemia, acute lymphocytic leukemia, acute myeloid leukemia, B-cell prolymphocytic leukemia, hairy cell leukemia, chronic lymphocytic leukemia, chronic myeloid leukemia, chronic T-cell lymphocytic leukemia, eosinophilic leukemia, Li-Fraumeni Syndrome, liver cancer, lung cancer, Hodgkin lymphoma, Non-Hodgkin lymphoma, Lynch Syndrome, mastocytosis, medulloblastoma, melanoma, meningioma, mesothelioma, Muir-Torre Syndrome, multiple endocrine neoplasia type 1, multiple endocrine neoplasia type 2, multiple myeloma, myelodysplastic syndromes, MYH-associated polyposis, nasal cavity and paranasal sinus cancer, nasopharyngeal cancer, neuroblastoma, neuroendocrine tumors, neurofibromatosis type 1, neurofibromatosis type 2, nevoid basal cell carcinoma syndrome, oral and oropharyngeal cancer, osteosarcoma, ovarian cancer, pancreatic cancer, parathyroid cancer, penile cancer, Peutz-Jeghers Syndrome, pituitary gland tumors, pleuropulmonary blastoma, prostate cancer, retinoblastoma, rhabdomyosarcoma, salivary gland cancer, sarcoma, alveolar soft part and cardiac sarcoma, Kaposi sarcoma, skin cancer, small bowel cancer, stomach cancer, testicular cancer, thymoma, thyroid cancer, tuberous sclerosis syndrome, Turcot Syndrome, unknown primary, uterine cancer, vaginal cancer, Von Hippel-Lindau Syndrome, Wilms tumors, and Xeroderma pigmentosum.

[0102] In embodiments, methods herein include the selection of a treatment, adjustment of one or more doses of a treatment, and / or the termination of a treatment. In such embodiments, annotation of one or more cells originating from a subject provide cell signatures that aid in diagnosis and / or selection of a cognate treatment. This includes, in embodiments, one or more chemical-genetic models or algorithms which aid the selection of one or more treatments for the identified mutants, for example with respect to malignancies, selection of one or more of a chemotherapy agent, radiation therapy, and surgery. In embodiments, methods herein include adjusting one or more doses of a treatment, for example, in subjects is undergoing treatment for a disease, e.g., such as an immunotherapy agent, chemotherapy regimen, radiation therapy, or surgery for cancer. In embodiments, methods include the termination of a treatment, for example, where a subject’s cell signature indicates that a disease is treatment refractory, or has a disease that has developed drug resistance, e.g, such as the identification of drug resistance mutant cells among a sample. In embodiments, method herein identify that a subject is treatment refractory, or has a disease that has developed drug resistance (e.g., from the identification of one or more mutations).

[0103] In embodiments, the subject of methods herein has one or more of or is suspected of having one or more of lymphopenia, cancer, immune deficiency, autoimmune disease, skeletal dysplasia, a bone marrow failure syndrome, genetic disorder impacting the immune system, cardiac failure, neural disorders, immunodeficiency, blood disorders (e.g., thalassemia, anemias, sickle cell disease), heart disease, liver disease, multiple sclerosis, muscular dystrophy, skin and tissue regeneration, spinal cord degeneration, trauma, stroke, neurodegenerative diseases (e.g., Alzheimer, Huntington’s, dementia, down’s syndrome, Parkinson, etc.), metabolic disorder, hematopoietic stem cell transplant (HPSCT), i.e., administration of healthy hematopoietic stem cells to patients with dysfunctional or depleted bone marrow, thrombocytopenia, or cancer.

[0104] In embodiments, the methods described herein inform the selection of a treatment, adjustment of one or more doses of a treatment, and / or the termination of a treatment, where the treatment includes any treatment linked to a chemical-genetic effect or phenotypic difference that relates to or correlates with one or more identified mutations. The treatment, in embodiments, includes fusion proteins, gene therapies, cell-based therapies, immunotherapies, chemotherapy agents, antibody-based agents, targeted radiation therapy, surgery, and the like.

[0105] Libraries and Models of Cellular Data

[0106] In aspects, the disclosure provides libraries and / or models of cellular data comprising the transcriptome profile and genotype sequence of one or more cells or tissues. In embodiments, the models of cellular data include one or more single cell genotypes matched to one or more disease states, and data relating the locations of one or more mutations and / or single nucleotide polymorphisms in the one or more single cell genotypes to one or more disease states.

[0107] Mutational information, in embodiments, includes substitutions, variants, deletions, duplications, insertions, translocations, or indels; single nucleotide polymorphisms SNPs are distinguished in that these may be polymorphisms that do not necessarily relate to disease, mutation (silent), or a change in phenotype. In embodiments, the library of cellular data includes one or more phylogenetic lineage trees (e.g., for one or more individual genes).

[0108] In embodiments, the library of cellular data includes information on how the one or more mutations and / or single nucleotide polymorphisms changes as a function of time (e.g., transcriptome dynamics and variability). In embodiments, this data relates to differences in time, treatment, age of subject, disease progression, etc.

[0109] In embodiments, the library is then used as an input for both short read RNA-seq and genotyping. In embodiments, genotyping targets are enriched by hybrid capture using probes that target full-length cDNA from select transcripts. In embodiments, the single cell genotyping comprises short read sequencing. In embodiments, the single cell genotyping comprises long read sequencing. In embodiments, long read sequence data retain the cell barcode, and thus can be is used to map variant status from transcripts onto cells identified in the gene expression data.

[0110] In embodiments, the library of cellular data comprises one or more machine-learning models and / or outputs, as described herein, useful for annotating single cell genotypes or transcription profiles. The machine-learning model of output includes, in embodiments, one or more of UMAP (e.g., with or without densMAP), t-SNE, PCA, correlation activity analysis (e.g., related / -test and / or v-scores), leave-one-out classifier analysis, trained logistic regression classifier, nearest-neighbor analysis or trained atrained nearest-neighbor classifier, confusion plot, differential expression analysis (e.g., DEseq2).

[0111] In embodiments, the library of cellular data comprises non-transitory data stored on one or more computing systems. Illustrative computing system include one or more of a computing device, database, server, laptop, tablet, smartphone, etc., which is able to store and / or retrieve data, and / or execute machine-readable code. In embodiments, the computing system comprises one or more computing devices, e.g., linked together in a network, virtual machine, or configured to function together via a cloud computing network. The computing system, in embodiments, includes internet-capability, or is otherwise able to retrieve or manipulate data from an internetbased or cloud-based storage. In embodiments, the computer system (e.g., comprising one or more computing devices) includes a network interface. In embodiments, the network interface interconnects the system computing devices within the system with each other, as well as optional external systems and devices, through one or more communication networks. In embodiments, the network interface optionally provides communication through a network communication module via the Internet, one or more local area networks (LANs), one or more wide area networks (WANs), other types of networks, or a combination of such networks.

[0112] In embodiments, one or more of the methods and / or elements of data described herein are stored in one or more non-transitory memory devices, e.g., as part of a computing system and / or one or more computing devices. In embodiments, the memory devices are configured to store, retrieve, and / or execute a set of instructions for performing one or more functions described herein. In embodiments, modules, data, or programs (e.g., sets of instructions) need not be implemented as separate software programs, procedures, datasets, or modules, and thus various subsets of these modules and data may be combined or otherwise re-arranged in various implementations. In implementations, non-persistent memory optionally stores a subset of the modules and data structures identified herein. Further, in embodiments, the memory stores additional modules and data structures not described herein, e.g., necessary to retrieve data, open program s / data, or otherwise execute programs / algorithms for manipulation or use of data or performance of methods described herein.

[0113] In embodiments, the library of cellular data comprises or is associated with an Application Program Interface (API) or Graphical User Interface (GPI) useful for retrieving, opening, and / or using the library of cellular data. Persons skilled in the art, with the benefit of this disclosure in its entirety, will be aware of the various forms the library of cellular data can take on, how to store, retrieve and use it, as well as the suitable hardware and software for doing so.

[0114] Without further elaboration, it is believed that one skilled in the art can, based on the above description, utilize the present disclosure to its fullest extent. The following specific embodiments are, therefore, to be construed as merely illustrative, and not limiting of the remainder of the disclosure in any way whatsoever.

[0115] EXAMPLES

[0116] Example 1; Single Cell Genotyping

[0117] This single cell genotyping approach was demonstrated in a mixed sample where genotypes were correctly assigned to known cells, with genotype well correlated to expression pattern. For this a cell line from a patient with a hematological proliferative disorder harboring a driver mutation in JAK2 and a matched cell line wild-type at the JAK2 locus was used. Mutation calls were also confirmed in parallel using a published nested PCR method (AmpliSeq, e.g., performed substantially as described in Van Egeren et al., Cell Siem Cell. 2021); all confident mutation call agreed between the two methods across overlapping cell barcodes. Finally, 8 hematological patient samples (8) were successfully profiled to identify the mutant carrying subpopulations. Greater than 99.5% of genotypes were concordant between the present hybridization-based method and the amplification-based method, and simultaneous profiling JAK2 and several other driver genes from individual cells was achieved. Example 2; Hybrid Capture Approach for Efficient Single Cell Genotyping

[0118] Many diseases, including several hematopoietic cancers, are driven by a small number of somatic mutations acquired by progenitor cells. In this Example, high-resolution transcriptional profiles of cells in healthy and disease populations were used for drug discovery. Without being bound to a particular theory, RNAseq + genotype data from single cells allowed for studies of mutant subpopulations of cells. This Example describes the use of a combination of single-cell RNA-seq with genotyping information obtained from hybrid capture and long-read sequencing for examining transcriptional profiles. Mutant cells from donor PBMC samples were identified and characterized and validated with normal and mutant control cells.

[0119] Previously described single-cell genotyping methods are based on the use of nested PCR. The nested PCR approach relied on design and validation of a complex set primers for each mutation to be targeted and requires a large number of PCR cycles (—50). Large numbers of PCR cycles lead to a measurable rate of crossover product errors, which can result in misassignment of genotype data.

[0120] Without being bound to a particular theory, this Example describes an approach to address these challenges by using an easily scalable hybrid capture method and long read sequencing to genotype multiple loci simultaneously and sequence transcriptomes in single cells while largely eliminating error from PCR.

[0121] FIG. 1 illustrates the results of the hybrid capture panel design in comparison between a 17-gene panel and a 2-gene panel. The table in FIG. 1 shows the hybrid capture gene panel metrics. The 2-gene panel showed sufficient coverage for genotyping low expressed genes. Both the 17- gene panel and 8-gene panel did not have enough coverage for low expressed genes. FIG. 1 shows that imbalanced gene expression in panel limits genotyping of lowly-expressed genes.

[0122] For accurate genotyping, callable read alignments were defined by the neighborhood quality standard, requiring an exact sequence match at the position of the mutation and the flanking lObp. For a cell to be called mutant, 2X as many Unique Molecular Identifiers (UMIs) with mutant calls vs wildtype calls were required.

[0123] FIG. 2A-2C illustrates the results of single cell genotyping applied to cell lines with known mutation status. In FIG. 2A, the illustration shows a comparison of control cell line genotypes UKE-1 and HL60 on PacBio. UKE-1 was mutant at the target gene locus; HL60 was wild type. Uniform Manifold Approximation and Projection (UMAPs) of expression data from controls were annotated with genotype calls from hybrid selected + PacBio data. In FIG. 2B, the illustration shows a comparison of mixed UKE-1 and HL60 (1 :1) genotypes from nested PCR seq. Nested PCR enabled distinct clustering of cells based on genotype with some cells ambiguously called due to PCR error. In FIG. 2C, the illustration shows a comparison of mixed UKE-1 and HL60 (1: 1) genotypes from on PacBio long read sequencing. Identical genotype-based clustering was observed using long read single-cell genotyping, with reduced ambiguity and more cells called overall.

[0124] FIG. 3 illustrates the results of single cell genotyping results compared across various methods. FIG. 3 shows the concordance demonstrated between genotype methods, called cells nested PCR and called cells long reads.

[0125] FIG. 4 illustrates the results of single cell genotyping results from nested PCR vs long read hybrid capture. FIG. 4 showed results of mutant cells resolved in patient CD34+ scRNA-seq data from PacBio. The disease sample showed clear subpopulation of cells with mutant genotype. Without being bound to a particular theory, the long-read hybrid capture was able to determine the cell genotype with more accuracy compared to the nested PCR as shown in FIG. 4, where the long read hybrid capture determined all mutant cells while the Nested PCR reads were only able to determine some mutant cells as ambiguous.

[0126] DEFINITIONS

[0127] The following definitions are used in connection with the disclosure disclosed herein. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood to one of skill in the art to which this disclosure belongs.

[0128] In embodiments, as used herein, “a,” “an,” or “the” refer to one or more than one.

[0129] Further, the term “about,” in embodiments, when used in connection with a referenced numeric indication means the referenced numeric indication plus or minus up to 10% of that referenced numeric indication. For example, the language “about 50” covers the range of 45 to 55.

[0130] As used herein, the word “include,” and its variants, is intended to be non-limiting, such that recitation of items in a list is not to the exclusion of other like items that may also be useful in the compositions and methods of this technology. Similarly, the terms “can” and “may” and their variants are intended to be non-limiting, such that recitation that an embodiment can or may comprise certain elements or features does not exclude other embodiments of the present technology that do not contain those elements or features.

[0131] Although the open-ended term “comprising,” as a synonym of terms such as including, containing, or having, is used herein to describe and claim the disclosure, the present disclosure, or embodiments thereof, may alternatively be described using alternative terms such as “consisting of’ or “consisting essentially of.”

[0132] In embodiments, as used herein, the words “preferred” and “preferably” refer to embodiments of the technology that afford certain benefits, under certain circumstances. However, other embodiments may also be preferred, under the same or other circumstances. Furthermore, the recitation of one or more preferred embodiments does not imply that other embodiments are not useful and is not intended to exclude other embodiments from the scope of the technology.

[0133] References

[0134] Becht et al., “Dimensionality reduction for visualizing single-cell data using UMAP,” Nature Biotechnology, Vol. 37, No. 1, 2019: pp. 38-45.

[0135] Blondel, et al. (2008), “Fast unfolding of communities in large networks,” Journal of statistical mechanics: theory and experiment, 2008(10), Pl 0008.

[0136] Campello, et al., (2015), “Hierarchical density estimates for data clustering, visualization, and outlier detection,” ACM Transactions on Knowledge Discovery from Data (TKDD), Vol. 10, No. 1.

[0137] Cunningham, '''Dimension Reduction,” University College Dublin, Technical Report, (2007), UCD-CSI-2007-7.

[0138] ENCODE Project Consortium, “An integrated encyclopedia of DNA elements in the human genome,” Nature, Vol. 489, No. 7414, 2012: pp. 57-74.

[0139] Fodor, “A survey of dimension reduction techniques,” Center for Applied Scientific Computing, Lawrence Livermore National, Technical Report, (2002), UCRL-ID-148494. Hafner et al., “Barcoded cDNA library preparation for small RNA profiling by next-generation sequencing,” Methods, Vol. 58, No. 2, 2012: pp. 164-70.

[0140] Maston et al., “Transcriptional regulatory elements in the human genome,” Annu Rev Genomics Hum Genet, Vol. 7, 2006: pp. 29-59.

[0141] Mclnnes et al. (2018) “UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction,” ArXiv e-prints, 1802.03426.

[0142] Mikheenko et al., “Sequencing of individual barcoded cDNAs using Pacific Biosciences and Oxford Nanopore Technologies reveals platform-specific error patterns,” Genome Res., Vol. 32, 2022: 726-37.

[0143] Narayan et al., “Assessing single-cell transcriptomic variability through density-preserving data visualization,” Nature Biotechnology, Vol. 39, No. 6, 2021: pp. 765-74.

[0144] Padmaja and Vishnuvardhan, “2016 IEEE 6th International Conference on Advanced Computing (IACC),” 2016: pp. 31-34, ISBN 978-1-4673-8286-1.

[0145] Schweyen et al., “Detection and Removal of PCR Duplicates in Population Genomic ddRAD Studies by Addition of a Degenerate Base Region (DBR) in Sequencing Adaptors,” Biol. Bull., Vol. 2272014, 2014: pp. 146-60.

[0146] Van Egeren et al., “Reconstructing the Lineage Histories and Differentiation Trajectories of Individual Cancer Cells in Myeloproliferative Neoplasms,” Cell Stem Cell. 2021 Vol. 28, No. 3, 221 : pp. 514-23.

[0147] Wang and Zha, “Adaptive Manifold Learning,” Advances in Neural Information Processing Systems, Vol. 17, 2004: pp. 1-8.

[0148] Wang et al., “Nanopore sequencing technology, bioinformatics, and applications,” Nature Biotechnology, Vol. 39, 2021: pp. 1348-65.

[0149] Xu et al., “Design of 240,000 orthogonal 25mer DNA barcode probes,” PNAS, Vol. 106, No. 7, 2009: pp. 2289-94.

[0150] Zahorian and Hu, “Nonlinear Dimensionality Reduction Methods for Use with Automatic Speech Recognition,” Speech Technologies, (2011), ISBN 978-953-307-996-7. All references cited herein are incorporated herein by reference in their entirety and for all purposes to the same extent as if each individual publication or patent or patent application was specifically and individually indicated to be incorporated by reference in its entirety for all purposes.

Claims

CLAIMS1. A method for single cell genotyping comprising: generating a single cell gene expression profile from a single cell in the sample of cell and / or a clonal cell population originating from the sample of cells, wherein the single cell gene expression profile comprises sequencing data from a cDNA library and / or gene capture; and performing a machine-learning process using the sequencing data from the cDNA library or the gene capture to generate a single cell gene expression model, wherein the single cell gene expression model provides information for annotating a single genotype.

2. The method of claim 1, further comprising obtaining a sample of cells3. The method of claim 2, further comprising expanding one or more clonal cell populations from one or more cells in the sample of cells.

4. The method of any one of claims 1-3, wherein generating the single cell gene expression profile comprises generating the cDNA library from the single cell.

5. The method of claim 4, wherein generating the cDNA library comprises performing reverse transcriptase PCR (RT-PCR) from mRNA.

6. The method of claim 4 or 5, wherein generating the cDNA library comprises barcoding the cDNA.

7. The method of claim 6, wherein barcoding the cDNA further comprises performing one or more of fragmentation, end repairing, A-tailing, and adaptor ligation, optionally performing each of fragmentation, end repairing, A-tailing, and adaptor ligation.

8. The method of claim 6 or 7, further comprising amplifying the cDNA prior to and / or during barcoding.

9. The method of any one of claims 1-8, wherein generating the single cell gene expression profile comprises sequencing the cDNA library, optionally a barcoded cDNA library.

10. The method of claim 9, wherein sequencing the barcoded cDNA library comprises indexing the barcoded cDNA library using sample index PCR, wherein each output provides a barcoded DNA sequence.

11. The method of claim 9 or 10, wherein sequencing the cDNA library comprises performing nested PCR.

12. The method of any one of claims 9-11, wherein sequencing the cDNA library comprises performing short read sequencing, long read sequencing, or both.

13. The method of any one of claims 9-12, wherein sequencing the cDNA library comprises performing amplicon sequencing.

14. The method of any one of claims 9-13, wherein sequencing the cDNA library comprises performing one or more of single-cell RNA sequencing (scRNA-seq), scTag-seq, singlecell transposase-accessible chromatin sequencing (scATAC-seq), CyTOF / SCoP, E- MS / Abseq, miRNA-seq, CITE-seq, and combinations thereof.

15. The method of any one of claims 1-14, further comprising performing whole genome sequencing (WGS) to generate genomic sequencing data, optionally from a single cell from a clonal cell population.

16. The method of claim 15, further comprising generating one or more lineage trees of one or more genes of interest from the genomic sequencing data.

17. The method of claim 15, further comprising refining the single cell gene expression model using the sequencing data from the cDNA library or the gene capture.

18. The method of any one of claims 1-17, wherein generating the single cell gene expression profile further comprises performing single cell gene capture to generate gene capture sequence data of one or more genes of interest.

19. The method of claim 18, wherein the gene capture is performed with the cDNA.

20. The method of claim 18 or 19, wherein the gene capture comprises hybridizing the cDNA with one or more hybridization probes that target one or more of the genes of interest.

21. The method of any one of claims 18-20, wherein the gene capture comprises ligating the cDNA with one or more adaptors.

22. The method of claim 19 or 20, wherein the one or more hybridization probes and / or one or more adaptors are chemically-modified.

23. The method of claim 22, wherein the one or more chemical -modified hybridization probes and / or one or more chemical-modified adaptors is biotinylated or streptavidin-tagged.

24. The method of any one of claims 18-23, wherein performing the gene capture comprises enriching one or more of the genes of interest using magnetic beads, wherein the magnetic beads bind the one or more hybridizing probes and / or adaptors.

25. The method of claim 24, wherein the magnetic beads bind the one or more hybridizing probes and / or adaptors via a biotin-streptavidin interaction.

26. The method of any one of claims 18-25, wherein generating the gene capture sequence data comprises performing sample index PCR and / or nested PCR.

27. The method of any one of claims 18-26, wherein generating the gene capture sequence data comprises sequencing one or more of the genes of interest using short read sequencing, long read sequencing, or both.

28. The method of any one of claims 18-27, wherein generating the gene capture sequence data comprises sequencing one or more of the genes of interest using amplicon sequencing.

29. The method of any one of claims 18-28, wherein generating the gene capture sequence data comprises performing one or more of single-cell RNA sequencing (scRNA-seq), scTag- seq, single-cell transposase-accessible chromatin sequencing (scATAC-seq), CyTOF / SCoP, E-MS / Abseq, miRNA-seq, CITE-seq, and combinations thereof.

30. The method of any one of claims 1-29, wherein the machine-learning process comprises manifold learning and / or dimensionality reduction.

31. The method of claim 30, wherein manifold learning comprises performing one or more of uniform manifold approximation and projection (UMAP), locally linear embedding, local linear isometric mapping (ISOMAP), t-distributed stochastic neighbor embedding (t-SNE), or Potential of Heat-Diffusion for Affinity Based Trajectory Embedding (PHATE).

32. The method of claim 30, wherein dimensionality reduction comprises performing one or more of uniform manifold approximation and projection (UMAP) algorithm, principal component analysis (PCA), random projection, independent component analysis, feature selection method, factor analysis, Sammon mapping, curvilinear components analysis, stochastic neighbor embedding (SNE), local linear isometric mapping (ISOMAP), maximum variance unfolding, locally linear embedding, t-distributed stochastic neighbor embedding (t-SNE), non-negative matrix factorization, kernel principal component analysis, graph-based kernel principal component analysis, linear discriminant analysis, a generalized discriminant analysis, LargeVis, Laplacian Eigenmap, and Fisher's linear discriminant analysis.

33. The method of any one of claims 1-29, wherein the machine-learning process comprising clustering.

34. The method of claim 33, wherein the clustering comprises performing HDBSCAN and / or Louvain community detection.

35. The method of any one of claims 1-34, further comprising annotating one or more single cell genotypes using the single cell gene expression model.

36. The method of claim 35, wherein the annotating further comprises using the single cell gene expression model to identify one or more mutations and / or single nucleotide polymorphisms.

37. The method of claim 36, wherein the one or more mutations comprise one or more substitutions, variants, deletions, duplications, insertions, translocations, or indels.

38. The method of claim 36 or 37, wherein the one or more mutations comprise somatic mutations and / or germ line mutations.

39. The method of any one of claims 36-38, wherein the one or more mutations are located in one or more of non-protein coding RNAs, intronic sequences, promoters, enhancers, silencers, insulators, structural elements of chromosomes, transcription-factor binding elements, histone-binding regions, telomeres, and satellite DNA.

40. The method of any one of claims 36-39, wherein the one or more mutations cause or are correlated with one or more changes in protein function.

41. The method of claim 35, wherein annotating the single cell genotype comprises classifying the single cell into one or more categories based on the identification of one or more mutations.

42. The method of claim 41, wherein the one or more categories comprises a healthy (wildtype) category and a diseased (abnormal) category.

43. The method of any one of claims 1-42, wherein the method improves detection of a gene signature compared to detection of a gene signature using single cell genotyping alone.

44. The method of any one of claims 1-43, wherein the method improves detection of a gene signature compared to detection of a gene signature using single cell gene expression alone.

45. A method for single cell genotyping comprising:1) obtaining a sample of cells and expanding one or more clonal cell populations from the sample of cells;2) generating a single cell gene expression profile from a single cell in the sample of cell and / or from the one or more clonal cell populations, wherein the single cell gene expression profile comprises sequencing data from a cDNA library and gene capture;3) generating genomic sequencing data by performing whole genome sequencing (WGS) using a single cell from the one or more clonal cell populations and generating one or more lineage trees using the genomic sequencing data;4) performing a machine-learning process using one or more of the sequencing data from the cDNA library, gene capture, and the genomic sequencing data to generate a single cell gene expression model;5) refining the single cell gene expression model using one or more of the sequencing data from the cDNA library, gene capture, and the genomic sequencing data; and6) annotating one or more single cell genotypes using the refined single cell gene expression model, wherein annotating comprises identifying one or more mutations in the single cell and classifying the single cell into one or more categories as a function of the identifying.

46. A method for measuring disease progression comprising performing the method of any one of claims 1-45, wherein at least a first single cell genotype is annotated at a first timepointand at least a second single cell genotype is annotated at a second timepoint, wherein a difference in one or more mutations between the first timepoint and second timepoint indicates a progression in disease state.

47. The method of claim 46, wherein the single cell originates from a blood draw, biopsy, or a clonal cell population derived therefrom, optionally wherein the blood draw comprises circulating tumor cells, or the biopsy is a tumor biopsy.

48. The method of claim 46 or 47, wherein a time between the first timepoint and the second timepoint is about or at least about 1 day, about or at least about 2 days, about or at least about 3 days, about or at least about 1 week, about or at least about 2 weeks, about or at least about 3 weeks, about or at least about 1 month, about or at least about 2 months, about or at least about 6 months, about or at least about 1 year, about or at least about 2 years, or about or at least about 5 years.

49. The method of any one of claims 46-48, further comprising stratifying a subject as a function of the progression in disease state, optionally wherein stratifying comprising classifying the subject into one or more categories based on a change in the disease state.

50. The method of claim 49, wherein the one or more categories are or relate to one or more diseases, optionally comprising a solid tumor, hematological cancer, acquired disease, congenital disease, hematopoietic disease, non-hematopoietic disease, disease affecting bone marrow, neurons, astrocytes, oligodendrocytes, cardiomyocytes, skeletal muscle cells, hepatocytes, pancreatic P cells, and lung epithelial cells.

51. A method for treating a disease in a subject in need thereof comprising performing the method of any one of claims 1-50, wherein the method informs the selection of a treatment, adjustment of one or more doses of a treatment, and / or the termination of a treatment.

52. The method of claim 51, further comprising performing a correlation activity analysis to identify a compound that induces a transcriptional signature.

53. The method of claim 51 or 52, further comprising performing a leave-one-out classifier analysis to identify a compound-specific differential transcriptional signature.

54. The method of any one of claims 51-53, further comprising performing a nearest-neighbor analysis to assess the consistency and / or reproducibility of compound-specific differential transcriptional signatures.

55. The method of any one of claims 51-54, further comprising performing a differential expression analysis to determine which genes are differentially regulated by one or more compounds.

56. The method of any one of claims 51-55, wherein the disease is or relates to one or more of a solid tumor, hematological cancer, acquired disease, congenital disease, hematopoietic disease, non-hematopoietic disease, disease affecting bone marrow, neurons, astrocytes, oligodendrocytes, cardiomyocytes, skeletal muscle cells, hepatocytes, pancreatic 0 cells, and lung epithelial cells.

57. The method of claim 56, wherein the disease is a solid tumor or hematological cancer.

58. The method of any one of claims 51-57, wherein a single cell genotype originates from a blood draw, biopsy, or clonal cell population obtained from the subject, optionally wherein the blood draw comprises circulating tumor cells, or the biopsy is a tumor biopsy.

59. The method of any one of claims 51-58, wherein the subject is undergoing treatment for cancer.

60. The method of any one of claims 51-59, wherein the subject is treatment refractory or has a disease that has developed drug resistance, and / or wherein the method identifies that the subject is treatment refractory or has a disease that has developed drug resistance.61 . The method of any one of claims 51-60, wherein the treatment comprises one or more of an immunotherapy, chemotherapy agent, radiation therapy, and surgery.

62. A library of cellular data generated from the method of any one or claims 1-50, wherein the library of cellular data comprises one or more of: i) one or more single cell genotypes matched to one or more disease states, and / or data relating to the identities of one or more mutations and / or single nucleotide polymorphisms in one or more single cell genotypes, and / or data relating the one or more mutations to one or more disease states; ii) one or more mutations and / or single nucleotide polymorphisms that change as a function of time and / or kinetics of mutation change; iii) one or more whole genome sequences; and iv) one or more machine-learning models and / or outputs for annotating single cell genotypes.

Images